Home >Miscellaneous>The Ayr to Kirkcudbright Road













The Ayr to Kirkcudbright Road

There is a certain degree of confusion about an old road running south from Ayr to Dalmellington and beyond. On the early Ordnance Survey maps it is shown as a Roman road as far as Cairnennock, south of Dalmellington though beyond that point it is called a pack road. On the Military Survey maps of the 1750’s it is shown as the road from Kirkcudbright to Ayr.

It is natural enough to think that if it is Roman it would have ran down towards the Solway coast, perhaps to a harbour near Kirkcudbright or to the Roman fort at Glenlochar near Castle Douglas. Yet the historian Chalmers, who was the first to talk of it being a Roman road (based on information provided by Joseph Train and others) does not say this. He has the road branching off the main Nithsdale road, passing the Doon of Tynron, then crossing the Shinnel at Stenhouse and running to Drumloff in Glencairn. It then followed the Cairn-Water, passed Conrig and then Altrie to Holm in Carsphairn parish. From there it ran across the Polwhat ridge towards Dalmellington. This seems to be the reason the early OS maps mark only the Dalmellington to Ayr stretch as Roman. Interestingly the Maxwelltown 1st edition 1” map shows a short stretch of “Roman road” near Holm (Nether Holm of Dalquhairn) and the 1853 6" map shows the line from Holm to Polwhat Rig.

This means that the “pack road” south of Dalmellington was seen as separate from the Roman road. Chalmers’ road would have approached Dalmellington from the east by Windy Standard and not from the south as the pack road does. It has to be said that nothing has been found between Dalmellington and Polwhat Rig that could be the road.

James MacDonald, an eminent archaeologist of his day was sceptical about the Roman provenance of the road and argued that between Dalmellington and Ayr it was identical to the road shown on the Military Survey and in fact an early “parish” road. He also looked at the route reported by Chalmers in Dumfriesshire and Kirkcudbrightshire and although he confirmed there was a road did not think it was Roman. However, he makes the interesting suggestion that it might have been a via regia mentioned in an early charter of Melrose Abbey that had not been identified at that time.

Other attempts have been made to trace this road and notably Clarke suggests that there could have been a Roman patrol track in this area although the course he gives is different to that given by Chalmers.

So far as the pack road goes, it seems clear enough that it was in use in the middle ages when the monks of Vaudey Abbey who owned the lands of Keresban (Carsphairn) used it to reach the Solway coast. At a slightly earlier period, Melrose had a grange here as well as at Dalmellington. The fact that they built roads elsewhere is suggestive.

We will look at the issues raised above under three headings:
- The road between Ayr and Cairnennock. This is the stretch marked as Roman on the OS maps and which MacDonald thought was a parish road;
- The various attempts to trace a Roman road running from the Nithsdale road (a well-established Roman road) through the Penpont and Moniave areas over towards Dalmellington from where it would continue as the Roman road above;
- The road south of Cairnennock which is marked as a pack road as far as Stroangassel, north of St John’s Town of Dalry but which then continued southward to Kirkcudbright though not marked as a pack road.

The “Roman Road” - Ayr to Dalmellington

Based on 1914 OS map. With thanks to Ordnance Survey. The roads shown in this and the maps below are from the Military Survey. The Ayr-Kirkcudbright route runs via Dalmellington, Carsphairn, New Galloway and Laurieston.

Although Chalmers (1) is vague about the course between Polwhat Rig and Dalmellington, a distance of 8 or 9 miles, he is clearer about the course between Dalmellington and Ayr. The course he gives is Polwhat Ridge, Dalmellington, bridleroad to Littlemill, Burnhead, Chapmeknowes, Pennason (Polnessan), Smithston, Cube, Boreland, Mains Hill, Causeway, Percluan Mill, Brae, Lindsayton, Cockhill, Whitestanes, Ayr. On the route just before Ayr, there was a place called the Foul Causeway and old people called the road the Pict's Road and others the Roman Way.

The course is marked on the early 1" OS maps (NLS site) and in general is fairly close to the A713 Ayr - Dalmellington road. An exception is immediately north of the town where it ran on the high ground a mile or so east of the present road up to near Polnessan.

In 1893, James MacDonald (2) published a paper that cast doubts on the road being Roman. He noted that the Ayr-Dalmellington road had been built in the late 1700's and had superseded an earlier road. This earlier road is shown on the maps of both Armstrong (1775) and the Military Survey (c.1750) and is so close to the line of the road detailed by Chalmers that it must be the "Roman" road. The entry for Dalmellington in the New Statistical Account refers to a section being dug up about 1830 on Burnhead farm - it was 10 or 11 feet wide with a row of large kerbs on each side, filled between with smaller stones.

To determine how old it was he examined a stretch on the farm at Smithston and found that it had kerbstones with a bottoming of stones which was covered with smaller stones and then gravel to the depth of 4 or 5 inches. He says that this does not conform to the way Roman roads were built and as it is close to a description given by Macadam on how roads were built prior to the turnpikes, he concludes that it is not Roman.

There are in fact a few discrepancies in MacDonald’s identification of the “Roman” road with that of the Military Survey. In his paper he reproduced Roy's road alongside the current OS map of the same route. Thus Roy's Sillyhole - Burnhead line is straighter than MacDonald suggests and Roy's map tends to confirm that from Burnhead the road took a relatively straight line as far as the Drongan road rather than leaving the Rankinston road a mile or so further on as shown on Armstrong - this would have called for a distinct change of alignment on Roy.

The road leading to Purclewan Mill

North of Smithston, Roy's road matches the track MacDonald shows which passes Newfield, skirts the modern road west of Boreland and runs to Mains. North of Mains the course is actually marked on the early OS maps as a Roman Road. Chalmers and MacDonald have it going through Purclewan Mill and along the attractive old road with its double hedgerows that can be seen today from the Dalrymple road but this would have shown up as a major deviation on Roy. It is more likely that it just ran straight, more or less following the realigned modern road to line up with the stretch of the A713 north of the Dalrymple road where MacDonald has it rejoin after his Purclewan Mill deviation.

A mile north of here, the early OS maps shows the Roman Road again. It started at the slight bend just after the industrial estate, went through Cockhill Wood and crossed the A713 at Ailsa Hospital to become the farm track on modern maps which runs past Braston. It then passed to the west of Castlehill and ran directly to the centre of Ayr.

Such discrepancies do not however negate MacDonald's arguments. There is hardly any doubt that before the turnpike to Dalmellington, there was only the one road and its course is shown accurately on the Military Survey map. If there was a Roman road it would have to be this road yet as he argues it does not show signs of Roman construction nor is there any tradition that it was Roman before Train and Chalmers suggested this when early antiquaries like Gordon, Maitland and Roy took note of such traditions elsewhere in Scotland.

In 1953, a field study group was set up by the Ayrshire Archaeological and Local History Society to study the route. (3) Various sections of the road were followed, viz. Castlehill to Bank farm, the county boundary towards Dalmellington, Dalmellington by Craigmark to Burnhead and Benquhat, Rankinston, Kerse Square up Dunstan Hill, Smithston to Newfield and Holehouse Junction and Perclewan Mill towards the Dalrymple and Coylton road.

They found further traces on Balgreen Farm and a possible ford over Boreland Burn although nothing was found over the Muck Water near Dalmellington. In several places they had difficulty in identifying the road. They concluded that although recent archaeological work (early 1950's) was supportive of the notion that a Roman road may have ran through the Ken and Doon Valleys, it could not be proved that this particular road was Roman.

In 1974, Newall and Lonie (4) traced an old road along part of this route. It appeared to be earlier than the "Roman" road which is still visible as a holloway north of Dalmellington but which is accompanied on stretches by a 7 metre wide mound just to the west of it. This mound is crossed by the holloway near Craigmark Hill (NS470080).

At Dunaskin Burn, there is a hollow with heavy kerbstones beside it that they suggest may be the stretch dug out in 1830. From this point, the "Roman" road would head towards Polnessan. The older road was traced northward past Kilmein Hill to the pass between Bow Hill and Ewe Hill along the line of the later proposed turnpike to Rankinston.

Allan Wilson (5) discusses a possible route north of this point. An older road diverges from the B730 turnpike at Drongan and crosses the Taiglum Burn at NS452182. From there, it probably headed to Stair, Tarbolton and Fail. Beyond the A77, Roy has a road on the same alignment running towards Irvine.

Although McDonald's arguments are powerful, there still remains some doubt about this road as shown by later researchers continuing to try to establish that it was Roman. Even if it was improved in statute labour times it could still be Roman, or as Wilson (5) says its construction could be Roman - this was by no means as uniform as is often said.

Nithsdale to Dalmellington

Tynron Doon on right - Shinnel Water  on left
Tynron Doon on right of picture - Shinnel Water on left

As said, Chalmers has the road branching off the main Nithsdale road, passing the Doon of Tynron, then crossing the Shinnel at Stenhouse and running to Drumloff in Glencairn. It then followed the Cairn Water, passed Conrig and then Altrie to Holm in Carsphairn parish. From there it ran across the Polwhat ridge towards Dalmellington. The Ordnance Survey show a mile or so of the road on the 1" Maxwellton sheet at Black Hill (4° 5’ W; 55° 16’ N) and its continuation west of Holm of Dalquhairn on the 1853 Kirkcudbrightshire 6" map (sheet 2). This is useful as it allows another four miles of the road to be identified. It is shown running north of Mid Hill of Greenhead, south-west of Windy Standard and Trostan Hill and directly onto Polwhat Rig. Its course is shown to about 1 kilometre south of Dun Hill.

Another writer, William Wilson (6), suggests a route up the Shinnel Water, roughly parallel to the Cairn Water but three miles to the east. Initially the route is the same as Chalmers to near Stenhouse but it then runs up the east side of the Shinnel by Craigturra, Auchengibbert and Killiewarren. It continues directly to Shinnelhead. In places it is identical to a “Deil’s” dyke.

John Shaw (7) casts doubt on both these routes and considers that although the remains are impressive in places they are in fact head dykes.

MacDonald (8) examined the stretch of Roman road marked on the Maxwellton 1" OS sheet (4° 5’ W; 55° 16’ N) and found it to be more of a path than a road although some minimal work had been done on it. However, he makes the interesting suggestion that the Glencairn road could be that mentioned in an early charter of Melrose Abbey. The road is very close to the Southern Upland Way and can be accessed from Polskeoch to the east (c.3 miles) or Lorg to the west (c.4 miles)

Reproduced from Sheet 31 (half-inch series), published 1914. With thanks to Ordnance Survey.

The route was looked at by Clarke in his paper Upper Nithsdale in Roman Times (9) where he refers to a road that can be traced from Colt Hill (NX698992) past Black Hill and Coranbrae Hill. This would make it identical to the road MacDonald examined, i.e. to the road marked on the Maxwellton map. Interestingly, it may have marked the county boundary here. Although its course across the valley of the Ken was not apparent it reappears on the Mid Hill of Greenhead and Windy Standard and from his description it is clear that this is the road shown on the early 6" OS map. Clarke noted that it was about 10 feet wide with some metalling on its surface which was covered with peat about one foot deep, and had cuttings and causeways in places. He suggested that it might be Roman in origin but used for cavalry patrols rather than standard troop movements. If there was a Roman road from the Nith it would have made much more sense for it to have gone through Moniave to either Carsphairn or Dalry to link up with a north-south route.

Allan Wilson (10) notes Clarke’s findings and discusses the possibility of a road through Carsphairn and Dalry over to the fort at Newton Stewart.

Newall and Lonie (11) examined what seems to have been in part the “Roman” road of the Maxwellton sheet both on the ground and on pre-forestation aerial photographs. It was traceable from Colt Hill to Trostan Hill to the east and from Coranbae Hill down to the farm of Carlae. As this western stretch is on a different alignment from that examined by Clarke it is likely to be a different road. To the east however, it could easily be a continuation of the “Roman” road examined by Clarke and MacDonald and is little more than a mile from Shinnelhead on the line suggested by William Wilson.

In summary we can see how unsatisfactory the evidence is for a Roman road running up either the Shinnel or Cairn Waters where the road or roads have been confused with head dykes, where separate tracks are involved and much of the putative route is so vague, especially between Polwhat Ridge and Dalmellington, a distance of 8 or 9 miles. The road first examined by MacDonald holds more promise but even here its origin and purpose is obscure.

The Pack Road

Based on 1914 OS maps. With thanks to Ordnance Survey

South of Dalmellington, two roads are shown on the Military Survey map. The westerly one split soon after leaving the town with one route to Barrbeth and one down to Carsphairn and Kirkcudbright. The easterly road is not shown on modern O.S. maps but appears on older editions. It ran over the Town's Common close to the present A713 before cutting down to join the previously mentioned road just west of Mossdale. From here it made directly for Bryan's Heights and Cairnennock close to the county boundary. Its course can be followed south of here on the older O.S. maps as the "Old Pack Road".

A mile or so beyond Cairnennock it has the same alignment as the A713 (the former turnpike) as far as Carsphairn. South of Carsphairn it runs by Bardennoch Hill to the deserted village of Polmaddy and then over Stroangassel Hill to merge with the modern road a mile further on. From here to Kirkcudbright it is either on the modern line or deviates only slightly from this. The main deviations are just north of Dalry and generally south of Laurieston though it is never far from the modern road. It is not marked as a pack road south of Stroangassel.

Near Cairnennock

Near to Cairnennock, the road has made features and kerbstones on one stretch though the work does not seem extensive in some places. It is noticeable how it holds to the high ground. At Bardennoch Hill near Carsphairn it has been used by tractors but reverts to its natural state further up the hill. Again work has been minimal. Sections exposed by tractor tracks indicate it may have had coverings of river pebbles but these seem to thin out further up the hill. Crosses (NMRS) have been incised on rock slabs near the pack road. These may date from the time that Vaudey Abbey held the lands or relate to pilgrims journeying to Whithorn.

The Pack Road at Bardennoch Hill Lower down the road has been used by tractors
Section lower down the hill Section higher up with far fewer pebbles

In the middle ages Thomas Colville who was lord of Dalmellington and Keresban (Carsphairn) leased the lands of Carsphairn to Melrose Abbey in about 1190. It seems that it was too remote for them for they let the land go in exchange for grazing rights in the Lammermuirs.

Colville then granted the lands to Vaudey Abbey in Lincolnshire. This too was Cistercian. There is a record of 1220 that they received permission from Henry III to import grain from Ireland which can only have been brought in from the coast, almost certainly by the pack road. Eventually the monks found the area too dangerous and relinquished their holdings back to Melrose (12, 13).

Given that the Military Survey notes it as the road between Ayr and Kirkcudbright, it was probably most in use when these two towns became well established in the middle ages.

Inn at Polmaddy The Pack Road just north of Polmaddy
The Inn at Polmaddy The Pack Road just north of Polmaddy

The well-known settlement of Polmaddy lies on the route. This was a fermtoun that went into decline in the late 1700’s perhaps as a result of the land being turned over to sheep farming. Although little is recognisable today a number of the buildings can be seen on the old 6”maps. There was an inn, right on the packhorse route. Polmaddy had a long history as Robert the Bruce is recorded as staying the night.

Based on 1935 1/4 inch OS map. With thanks to Ordnance Survey

As said, south of Stroangassel the road continued to Kirkcudbright. Apart from one or two tracks that seem the same as the Military Survey road, nothing is shown on the early OS maps which indicates that it fell out of use presumably when roads were built in the late 1700’s.

In earlier times it would have provided a route between Aeron and central Galloway, both of which were part of Rheged. At a later date it could have provided a route from the Anglian settlements in the New Galloway area up to their settlements in the vicinity of Maybole, and some researchers would place the toll point of Laicht Alpin near to Dunaskin, north of Dalmellington.

The only definite facts known about this road are that it was in use in the middle ages and may have been improved along parts of its route, probably by statute labour. It was also known as the route between Ayr and Kirkcudbright at the time of the Military Survey around 1750. It has suffered greatly from agricultural improvements, from mining and from afforestation and from being replaced by modern roads, and is now only traceable as a rough track in a few places.

As to the general problem as to why the road exists and when it first developed, several possibilities suggest themselves:
- that it was in fact Roman and being a natural communication route would have been used in the Dark Ages when Aeron was part of Rheged and later by the Angles to reach their settlements in Carrick. An alternative possibility has to be noted, however, which is that such a road would most likely have originated from Glenlochar and have ran on the east side of the Dee and only joined the Ayr - Kirkcudbright road somewhere near Carsphairn; i.e. it would have been a different road from the pack road south of Carsphairn but would have continued into Ayrshire as the "Roman" road (5)- this means that the pack road would have developed independently despite being noted on the Military Survey as a continuous road between Ayr and Kirkcudbright;
- that either the whole road or only the "pack road" had its origins in mediaeval times where Melrose Abbey had granges in Dalmellington and Carsphairn which would have required the transport of produce, as was also the case when Vaudey Abbey had Carsphairn and are said to have imported grain from the coast as well as transport   materials to the south;
- that the whole road developed as a trading route between Kirkcudbright and Ayr in the middle ages as well as being used by Melrose and Vaudey;
- that although it was in existence at least in mediaeval times (as evidenced by the monks) it could have been improved in some parishes under the statute labour system from the late 1600's onwards. This could explain why the road seems to have been well made in Ayrshire and less so further south.

Such speculations have their difficulties of course but they do suggest possible lines of enquiry that will prove or disprove any one hypothesis. Until then we will have to be satisfied with the little we do know about this road.

(1) George Chalmers, Caledonia, 1824, Vol iii, pps 236-7; 448-9
(2) James MacDonald, Notes on the "Roman" Roads of the One-inch Ordnance Map of Scotland, PSAS, vol.27, 1893, pp.417-43
(3) AANHS Collections, 2nd Series, Vol 3, 1955, pps 30-33
(4) Newall and Lonie, Discovery and Excavation in Scotland, 1974, p26
(5) Allan Wilson, Roman Penetration in Strathclyde South of the Antonine Wall -Part 1 - The Topographical Evidence, Glasgow Archaeological Journal, Vol.19, 1995, pps 1-30 (see pps.12 & 13)
(6) William Wilson, Tynron, Dumfriesshire from the Mists of Antiquity, and Verse, Dumfries, 1957
(7) John Shaw, Tynron Glen, 1995 - see also website
(8) James MacDonald, Notes on the "Roman" Roads of the One-inch Ordnance Map of Scotland, The Dumfresshire Roads, PSAS, vol.28, (1893-94), pp.298-320
(9) J Clarke, Upper Nithsdale and Westwards in Roman Times, TDGNHAS, 3rd series, 30, 1951 - 1952, pps 111-120
(10) Allan Wilson, Roman Penetration in West Dumfries and Galloway, TDGNHAS, A Field Survey, III, 64, 7 (see p.13)
(11) Newall and Lonie, Discovery and Excavation in Scotland, 1990, p.11
(12) Bill Blyth, Kars Castle, Carsphairn, p.4
(13) Carsphairn Heritage Group, Carsphairn, p.6